Understanding How Polymer Properties Control OPV DevicePerformance: Regioregularity, Swelling, and MorphologyOptimization Using Random Poly(3-butylthiophene-co-3-octylthiophene) PolymersAmy S. Ferreira,⊥,† Jordan C. Aguirre,⊥,† Selvam Subramaniyan,‡ Samson A. Jenekhe,‡

Sarah H. Tolbert,*,†,§,∥ and Benjamin J. Schwartz*,†,∥

†Department of Chemistry and Biochemistry, University of California, Los Angeles, Los Angeles, California 90095-1569, UnitedStates‡Departments of Chemical Engineering and Chemistry, University of Washington, Seattle, Washington 98195-1750, United States§Department of Materials Science and Engineering, University of California, Los Angeles, Los Angeles, California 90095, UnitedStates∥California NanoSystems Institute, University of California, Los Angeles, Los Angeles, California 90095, United States

*S Supporting Information

ABSTRACT: The performance of polymer:fullerene bulkheterojunction (BHJ) photovoltaics is highly sensitive to themorphology of the polymer within the active layer. To tunethis morphology, we constructed both blend-cast andsequentially processed BHJ devices from the fullerenederivative [6,6]-phenyl-C60-butyric acid methyl ester(PCBM), in combination with a series of random poly(3-butylthiophene-co-3-octylthiophene)s with different fractionsof each monomer, with the goal of controllably varying theaverage polymer side-chain length. What we found, however,was that the most important parameter for predicting deviceperformance across this series of polymers was theregioregularity of the particular synthetic batch of polymerused, not the average side-chain length. Moreover, we found that regioregularity affected device performance in different waysdepending on the processing route: lower regioregularity led to improved performance for sequentially processed devices, butwas detrimental to the performance of blend-cast devices. We argue that the reason for this anticorrelation is that regioregularityis the single most important determinant of the relative crystalline of the polymer. The relative crystalline fraction, in turn,determines the ability of the polymer to swell in the presence of solvents. Polymer swelling is key to BHJ formation via sequentialprocessing, but can lead to overly mixed systems using traditional blend-casting methods. As a result, we find that the bestperforming polymer for sequentially processed devices is the worst performer for blend-cast devices and vice versa, highlightingthe importance of using both processing methods when exploring new materials for use in BHJ photovoltaics.

■ INTRODUCTION

Organic photovoltaics (OPVs) composed of semiconductingpolymer donors have seen growing interest since efficiencies ofover 10% were reached.1,2 Most devices are composed ofblended systems with conjugated polymers as the primaryphotoabsorbers and fullerene derivatives, such as [6,6]-phenyl-C60-butyric acid methyl ester (PCBM), as the electronacceptors. For such systems, the photovoltaic efficiency ishighly correlated with the nanometer-scale morphology of theblended components.3 Due to low exciton diffusion lengths, thepolymer and fullerene must be mixed on a 10−20 nm lengthscale to enable excitons to be split into free carriers at thedonor/acceptor interface.4 However, the two components alsomust have separate domains to prevent recombination5 and to

allow the free carriers to travel to their respective electrodes.6,7

This type of morphology, known as a bulk-heterojunction(BHJ),8 can be difficult to create by traditional blend-casting(BC) methods in which the polymer and fullerene arecodissolved in solution and then cast into a film on aconductive substrate. This is because blend casting relies onspontaneous demixing of the two components during thedrying process, and often requires a cosolvent or solventadditive.9−11 Additional processing can also be performed afterfilm formation by blend casting, including thermal anneal-

Received: March 31, 2016Revised: August 26, 2016Published: August 29, 2016

Article

pubs.acs.org/JPCC

© 2016 American Chemical Society 22115 DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

ing12−14 and/or exposure to solvent atmosphere15,16 to furtherdemix the two phases. The final blend-cast film morphology ishighly sensitive to these processing conditions and oftenrequires an Edisonian, trial-and-error approach to optimize theextent of phase separation.17

An alternate approach to blend casting is sequentialprocessing (SqP), in which a pure polymer film is first castand dried, and then the fullerene layer is deposited on top ofthe polymer film in a second step.18−26 In SqP, the fullerenesincorporate into the amorphous regions of the polymer filmbecause the fullerene-casting solvent swells the underlyingpolymer film,21,26−32 creating a BHJ morphology that can besimilar to that formed through blend casting. Fullerenediffusion into the film can be further facilitated throughthermal- or solvent-annealing postdeposition pro-cesses.8,24,26,30,31 The SqP route tends to preserve the degreeof crystallinity and other structural aspects of the initiallydeposited pure polymer film in the final BHJ morphology thatis created, enabling fairly precise control over the BHJarchitecture.32−35

Although SqP is generally a more controllable processingmethod, the morphology created in both blend casting and SqPis highly sensitive to the propensity of the polymer tocrystallize.24 Many groups have looked at controlling therelative crystalline fraction of the polymer by tuning either thepolymer side chains or the polymer molecular weight, both ofwhich can alter the resulting BHJ architecture.36−38 We havepreviously shown that for poly(3-hexylthiophene) (P3HT), thedegree of crystallinity has the opposite effect on the efficiencyof devices fabricated by blend casting and SqP, with increasedrelative crystalline fractions improving the efficiency of blend-cast devices but decreasing the efficiency of sequentiallyprocessed devices.24

In this work, our initial goal was to further explore the extentto which the relative polymer crystallinity and paracrystallinitycontrol the performance of OPV devices fabricated usingdifferent processing methods. To controllably tune polymerproperties, we took advantage of a series of randomcopolymers, poly(3-butylthiophene-co-3-octylthiophene)s(RBO)s,39 in which the percentage of 3-butyl versus 3-octylside chains could be varied over a large range (Figure 1a). Wenote that random copolymers have been explored previously asa means to tune side-chain composition40,41 or to incorporateheavier selenium atoms as a means to tune the polymer’senergy levels.42,43 Our plan was to explore the differencesbetween blend casting and SqP by examining versions of thewell-studied P3HT system with a continuously tunable averageside chain length (Figure 1b). In this way, we planned to tunethe relative crystalline fraction and thus the BHJ morphologywithout any changes in the system energetics.What we found, however, was that BHJ morphology and

device performance were not dominantly determined by thepolymer side-chain composition, as we had expected, butinstead were primarily correlated with the regioregularity ofeach batch of polymer. The regioregularity resulted from thesynthetic details of each batch of polymer samples and variednonmonotonically across the range of samples. We note thatother groups have noted a relation between conjugatedpolymer regioregularity and the behavior of polymer/fullereneOPVs,44−50 so in this work our aim became to preciselydetermine the correlation between regioregularity, relativecrystalline fraction, and photovoltaic device performance. Wefind that the overall device performance and regioregularity

were oppositely correlated with efficiency when optimizingOPV devices fabricated by SqP and blend casting, assummarized in the TOC graphic. We argue that thisextraordinary sensitivity to regioregularity arises becauseregioregularity is the strongest predictor of the relative polymercrystalline fraction and in turn, polymer swelling. Swelling is anessential component in the formation of the BHJ morphology,particularly for fabrication via SqP.31 Overall, we show that thesteps taken to optimize BHJ formation with one type ofprocessing result in highly nonideal morphologies for the other,so that both types of processing conditions must be explored tofind the globally optimum device architecture for any given setof materials.

■ EXPERIMENTAL SECTIONWe refer to the individual RBO polymers we used in this workas RBOx, where x is the percent of butyl-thiophene monomersin the copolymer chain. We synthesized five random poly(3-butylthiophene-co-3-octylthiophene)39 co-polymers with differ-ent percentages of butyl side changes: 20% (RBO20), 34%(RBO34); 50% (RBO50); 61% (RBO61); and 72% (RBO72). Wemeasured the regioregularities of these five co-polymersthrough 1H NMR to be 91.8%, 91.3%, 94.0%, 92.9% and90.8%, respectively. The number-average molecular weight andpolydispersity index (PDI, also known as dispersity, D) for eachRBOx can be found in Table S1 of the Supporting Information(SI).We fabricated the active photovoltaic device layers in this

work using two basic processing conditions: “slow-dried”, inwhich the OPV film spin-coating is halted while the film is stillwet,39 and “fast-dried”, in which the active layer is spun untildry. All devices were fabricated on ITO substrates that weresequentially sonicated in baths of detergent, water, acetone, andisopropyl alcohol followed by UV ozone treatment for 10 min.Forty nanometers of poly(ethylenedioxythiophene):poly-(styrenesulfonic acid) (PEDOT:PSS Clevios PVP AI 4083)was deposited onto ITO to form the hole collection layer.For the blend-casting procedure, a 1:0.8 wt/wt

RBOx:PC71BM (Solenne, used as received) solution wasmade at a concentration of 20 mg/mL polymer/1,2

Figure 1. Characteristics of RBOX used in this study: (a) chemicalstructure, (b) average chain length, (c) polydispersity index (PDI), (d)regioregularity.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

22116

dichlorobenzene. This solution was deposited onto aPEDOT:PSS-coated substrate to produce films with a thicknessof 115 ± 5 nm. Slow-dried films were allowed to fully dry in aN2 atmosphere, then thermally annealed at 110 °C for 5 min.The fast-dried films were thermally annealed at 150 °C for 10min. We held the fullerene ratio constant in this work to focuson the effects of side-chain length and regioregularity, but wenote that polymers with lower regioregularity are able toaccommodate a higher degree of fullerene loading, as we havediscussed in previous work.24

Sequentially processed films were fabricated by depositingsolutions of the pure RBOx polymers onto PEDOT:PSS-coatedsubstrates at 1000 rpm for 60 s (fast-dried) or 30 s (slow-dried)at a concentration of 20 mg/mL in 1,2 dichlorobenzene.PC71BM was deposited onto the fully dry polymer layer by spincoating at 4000 rpm for 10 s from a 10 mg/mL solution indichloromethane. Both sequentially processed and blend-castprocessed films were thermally annealed at 150 °C for 10 min(fast-dried) or 110 °C for 5 min (slow-dried). Ten nanometersof Ca and 70 nm of Al were thermally evaporated (AngstromEngineering) as cathodes.Samples for GIWAXS were fabricated similarly to the

photovoltaic devices with the exceptions that Si was used asthe substrate (with a native ∼2 nm thick oxide overcoated by40 nm of PEDOT:PSS) instead of ITO on glass, and PC61BM(Nano-C) was substituted for PC71BM. Although changingfrom the C60 to C70 version of the fullerene might change thepolymer and/or fullerene crystallinities, the trends observedacross the RBOx series are expected to remain the same. Thethicknesses of all films were found to be comparable usingprofilometry (Dektak). GIWAXS data was taken on threesamples for each processing condition and corrected forfluctuations in beam intensity to ensure reproducibility. Thepolymer:fullerene ratio between blend-cast and sequentiallyprocessed films using a given RBOx donor was matched usingthe redissolved film absorption spectroscopy method detailedby Hawks et al.23

Photovoltaic performance was measured in an Ar atmosphereusing a Keithly 2400 source meter and AM-1.5-filtered lightfrom a xenon arc lamp equipped with a liquid light guide(Oriel). The incident light intensity on tested samples wasadjusted to be 100 mW/cm2 using a calibrated Si diode. Allphotovoltaic efficiencies were verified by comparing themeasured short-circuit current to the integrated externalquantum efficiency (EQE), as detailed in the SI.Grazing-incidence wide-angle X-ray scattering (GIWAXS)

measurements were performed at the Stanford SynchrotronRadiation Lightsource (SSRL) on beamline 11-3 using awavelength of 0.9742 Å. The diffraction patterns were collectedon a 2-D image plate with the detector 400 mm from thesample center. The beam spot was approximately 150 μm wide,and a helium chamber was used to reduce air scatter andimprove signal-to-noise. The WxDiff software package was usedto radially integrate scattering data. This was accomplished bytaking a fixed arc in chi of each diffractogram that includes bothin-plane and out-of-plane scattering. The subsequent 1-D datawas then corrected for the background silicon scattering. Sincescattering peaks lose intensity with increasing scattering angle, aq2 correction was performed to ensure that the (100) and (200)scattering intensities could be compared across the polymerseries. We note, however, that the peak shifts across thedifferent RBOx polymers were small enough that the q2

correction did not affect any of the results. The patterns

shown in the main text are results average from three separatefilms. Each diffractogram was fit to a series of Gaussian curves.The increase in the full width at half-maximum (fwhm) acrossthe overtones of the (100) scattering planes was used todetermine the g-factor, which is correlated with the paracrystal-line disorder.50 The subsequent g-factors for each separate filmwere averaged together for the correlation calculations. Theintegrated GIWAXS peaks provide information about therelative polymer crystallinity across the series, and the g-factorallows us to better understand correlations between regior-egularity and device performance with disorder in the polymer/fullerene films.Polymer swelling was measured in situ using a PS-1000

Semilab ellipsometric porosimeter; experiments following themethods outlined by Huttner et al.51 Pure polymer films weredeposited onto silicon substrates with a native oxide layer usingboth the fast-dried and slow-dried methods described above.Ellipsometric measurements were taken at the approximateBrewster angle of silicon, 75°. The data were fit using theSemilab Spectroscopic Ellipsometry Analyzer (SEA) software,using a multilayer model for the silicon substrate and the nativeoxide layer. The Cauchy model was used for the thin polymerfilm to fit the tan(Ψ) and cos(Δ) parameters directly measuredby the ellipsometer, as described in more detail in the SI. Nosignificant differences were found in the fitting parameterswhen anisotropy (i.e., birefringence) in the polymer films wasincluded in the model, so for simplicity, the data presented herewas fit assuming an isotropic (single n and k) polymer film.52

To minimize contributions from the polymer absorption, weused the 800−1000 nm region for our fits. The (unswelled)film thicknesses measured by ellipsometry were confirmed bydirect measurement with a Dektak profilometer. The swellingmeasurements used toluene vapor both because of itscompatibility with the porosimeter instrument and because itis a good solvent for the RBOX polymers that yields largevolume changes during swelling. Vapor pressures weremeasured as relative pressure (p/psat), using the saturationpressure psat for toluene at room temperature of 29 Torr. Therefractive indexes that we measured were validated usingreported values.19,53 Since the transparent region of theellipsometry data was used for fitting, changes in the refractiveindex with swelling and solvent incorporation were fully takeninto account. We note that we have shown in previous workthat only the amorphous regions of P3HT films swell to anysignificant degree, so that our swelling measurements comple-ment the GIWAXS data by providing additional information onnoncrystalline regions of the BHJ films.Although there are variations in the properties of each RBOx

polymer so that the behavior of photovoltaic devices fabricatedfrom these materials cannot be perfectly predicted by any givenpolymer property, what is of greatest interest is to determinethe degree of correlation between RBOx properties and deviceperformance. To determine correlations between differentparameters, we used the Pearson product−moment correlationcoefficient (r), given as54

=∑ − −

∑ − ∑ − r

x x y y

x x y y

( )( )

( ) ( )2 2(1)

where x and y are the data sets to be linearly correlated. Allcorrelation coefficients are given in Table 1. Although it ispossible that the correlation could be nonlinear, we found that

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

22117

the use of the linear Pearson correlation coefficient worked wellwith our limited data set of only five distinct copolymers.

■ RESULTS AND DISCUSSIONWe synthesized five RBOx random copolymers with butyl side-chain percentages of 20%, 34%, 50%, 61%, and 72% followingprocedures published previously.39 The PDI and regioregu-larity, which are both known to affect the polymer crystallinityand structure,3,45,55−57 also were measured39 and aresummarized in Figure 1c,d, respectively, as well as Table S1.Our original goal was to focus on average side chain length,rather than regioregularity or polydispersity. Although we triedto synthesize the different RBOx polymers in as similar amanner as possible, we were not able to precisely control thedegree of regioregularity of each synthetic batch of polymer,creating a nonmonotonic trend between average side chainlength and regioregularity. For this family of compounds,RBO50 has the highest regioregularity of the five RBOxpolymers studied. RBO20 and RBO72, by contrast, even thoughthey have drastically different monomer compositions, haveregioregularities that are quite similar. These latter twomaterials thus provide a means to isolate which factor ismore important for controlling the BHJ morphology and deviceperformance: regioregularity or average side chain length.A. Device Performance for the RBOx Series with

Different Processing Conditions. To begin our study, wefabricated both blend-cast and sequentially processedRBOx:PC71BM BHJ devices using the various processingconditions summarized above. The average power conversionefficiency and full average J−V curves for the best performingdevices are plotted in Figure 2. The full average J−V curves forthe processing conditions not shown in Figure 2 and otherperformance characteristics are given in the SI. It has beenshown previously that kinetics of film formation are improvedin blend-cast devices when slow-drying is used to increase theability of the polymer to reorganize and crystallize, rather thanbeing locked into a more amorphous, mixed architecture.58 Ourresults for blend-cast samples (red squares) fit with thisexpectation, with the slow-dried samples (panel c) out-performing the fast-dried samples (panel d). For both dryingconditions, the blend-cast (BC) devices with the best efficiencywere made using RBO50, whereas RBO20 and RBO72, whichhave identical regioregularity but very different side chains,produced devices with lower efficiencies. We calculated thecorrelation factor (eq 1) for the device efficiencies with variouscharacteristics of the different RBOx polymer batches, and theresults are summarized in Table 1 with additional details givenin the SI. Perhaps the most striking feature is that the overallefficiency trend for the blend-cast devices with both dryingconditions reflects the trend in regioregularity plotted in Figure1d: the correlation factors for both fast- and slow-dried blend-cast device efficiencies with polymer regioregularity are 0.75and 0.95, respectively, suggesting that regioregularity is indeed

an important factor in determining device efficiency. This is inagreement with previous work that shows that regioregularity inpolythiophenes leads to an increase in interchain stacking thatresults in higher time-of-flight mobilities and dark-currentdensities indicating better charge extraction from the device.44

The correlation between device efficiency and PDI also wascalculated (Table 1), but the correlation factors were far lower:there is no apparent trend with x, the average side-chaincomposition.The other striking feature of Figure 2 is that there is almost

complete anticorrelation between blend-cast and sequentiallyprocessed devices (black circles, panels c and d) in terms oftheir efficiencies. The sequentially processed devices show aninverse effect of regioregularity, with a high anticorrelationfactor (≤−0.9; see Table 1); the RBO50 material gives thelowest device performance for SqP. Additionally, the trend indevice performance with drying conditions is also reversedbetween blend casting and SqP, where fast-drying the initialpolymer film in SqP improves device performance. The factthat the optimal polymer material and processing conditionsgive precisely opposite trends for SqP and blend castingindicates that the overall kinetics of BHJ formation in these twofabrication processes is quite different.

B. Structural Studies of RBOx:PCBM BHJ Active Layers.To understand precisely what it is about the processingconditions that correlates or anticorrelates with deviceperformance for blend casting and sequential processing,respectively, we performed grazing incidence wide-angle X-rayscattering (GIWAXS) on all of the BHJ active layers to see ifpolymer crystallinity is at the root of these correlations.Crystalline regions are known to have a significant effect ondevice performance as they help create the energy cascade thatleads to charge separation and facilitate charge transport.Crystalline regions are also the key to creating the BHJmorphology via either SqP or blend casting because of the lowsolubility of fullerene in crystalline polymer. To simplify theresults for the sequentially processed active layers, we examinedthe structure of pure polymer films. This is because we haveshown previously for P3HT that PCBM intercalates primarily

Table 1. Summary of Pearson Product−Moment CorrelationCoefficients between RBOx/PC71BM Power ConversionEfficiency and RBOx Polymer Properties

active layer regioregularity PDI average side-chain length

slow dried SqP −0.96 −0.65 −0.25slow dried BC 0.95 0.35 −0.27fast dried SqP −0.94 −0.57 −0.28fast dried BC 0.75 0.57 0.62

Figure 2. J−V curves for slow-dried blend-cast (a) and fast-dried SqPfilms (b). The J-V curves for fast-dried blend-cast films and slow-driedSqP films are given in the SI. Photovoltaic device efficiency (%) forsequentially processed and blend-cast RBOx/PCBM films fabricatedwith the slow-dried method (c) and fast-dried method (d).

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

22118

into the amorphous polymer regions during SqP, so that thereis little change in X-ray-measurable polymer structure followingsequential fullerene deposition.23,24 The PCBM diffraction,however, overlaps the polymer diffraction and dramaticallycomplicates analysis of the scattering data, as seen by the largepeak at 1.4 nm−1 in Figure 3(b). For blend-cast samples, datafrom both pure polymer films and blend-cast active layers wereconsidered because the blended films can show very differentcrystalline properties than the pure polymer films.Since we have previously seen that polymer crystallinity plays

an important role in both blend casting and sequentialprocessing, we began by looking at the relative fraction ofpolymer crystallinity in both the lamellar (100) and π−πstacking (010) directions. To do this, we radially integrated our2-D GIWAXS diffractograms to obtain 1-D data, as shown inFigure 3 for fast-dried pure RBOx films (panel a) and slow-dried blend-cast films (panel b); integrated diffractograms forthe processing conditions not shown in Figure 3 can be foundin the SI. The resulting 1-D data was fit to a series of Gaussians,

where the peak area corresponds to the relative crystallinefraction across the series and the peak width correlates to aScherrer crystallite size or, more precisely, a domain correlationlength.50,59,60 The degree of paracrystalline disorder (g-factor)was also calculated from the spread of the full width at half-maximum (fwhm) of the lamellar peak, measured over multiplelattice overtones.50 Since this analysis relies on multiplediffraction overtones, g-factors could not be calculated for thepolymer (010) or fullerene peaks.The results for pure RBOx polymer films (black) and blend-

cast films with PCBM (red) are shown in Figure 3c−f. Fromthe data in Figure 3, we can readily see that the relative polymercrystalline fraction and crystallite size do not follow amonotonic trend with side-chain length, as we would haveexpected based on the standard model of increased disorderassociated with smaller side chains.36,37 This same trend inrelative crystalline fraction also was observed in DSCmeasurements on pure RBOx polymer films.39 This providesstrong evidence that the polymer crystalline properties, and in

Figure 3. Integrated GIWAXS diffractograms for fast-dried pure polymer (a) and slow-dried blend-cast films (b). Diffractograms for fast dried filmscan be found in the SI. Gaussian fits to GIWAXS lamellar (100) peak area (c), (100) crystallite size (d), (010) crystallite size (e), and paracrystallinedisorder (with one-sigma error bars) (f) for pure RBOx films (black curves, symbols) and RBOx/PCBM BHJs (red curves, symbols). IntegratedGIWAXS diffractograms for slow-dried pure polymer and fast-dried blend-cast films are given in the SI.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

22119

turn the BHJ morphology, are not a very strong function of theside-chain length but are instead mostly dependent on otherfactors.Looking at the trends in relative crystalline fraction for the

pure polymer films, we see very little change in the (010) peakarea across the RBOx series (Figure 3e), but striking changes inthe (100) peak area (Figure 3c). We thus use the (100) peakarea as a rough measure of relative fraction of RBOxcrystallinity. By this measure, for sequentially processed filmswith both drying conditions and for slow-dried blend-cast films,RBO50 has the highest relative crystalline fraction. The generaltrend is that pure polymer relative crystalline fraction correlateswell with polymer regioregularity, correlates somewhat withpolymer PDI, and does not correlate at all with chaincomposition, as summarized in Table 2. To ensure the accuracyof our measured relative crystalline fractions, we also performedthe same analysis using the (200) lamellar peaks and found thesame trend, as shown in more detail in the SI. On the otherhand, the extent of paracrystalline disorder, which is a measureof the long-range coherence of the crystalline packing,36,61−63

correlates well with the average side-chain length, with theshortest side-chain length associated with the largest disorder,as expected. Since order in the backbone of the polymer is seenin the relative crystalline fraction, this correlation most likelyreflects disorder within the side chains.In contrast to the correlations found with pure polymer films,

Table 2 shows that blend-cast RBOx films show almost nocorrelation between polymer regioregularity and relativecrystalline fraction. This is rather surprising given that it isgenerally accepted that both increasing side-chain length andincreasing polymer regioregularity improves polymer π−πoverlap by promoting interchain ordering.64,65 Clearly, thereis not a simple relationship between the synthetic parameters ofthe pure polymer and the polymer’s relative crystalline fractionin a blended system. Instead, a large variety of both polymerand fullerene characteristics play a role in the demixing thatoccurs during the drying (and thermal or solvent annealing) ofblend-cast films, so that there is no easy way to predict the finalBHJ morphology.17

To better assess how the relative fraction of polymercrystallinity is related to the BHJ morphology of blended films,we examined the correlation between device efficiency anddifferent GIWAXS parameters. Table 3 shows the correlation ofblend-cast and sequentially processed device efficiency withboth the blend-cast GIWAXS parameters and the pure polymerGIWAXS parameters across the RBOx polymer series. Thesequentially processed device efficiencies strongly anticorrelatewith the blend-cast device efficiencies, as expected based on ourprevious work showing that sequentially processed devices needmore amorphous regions for fullerene intercalation, whileblend-cast devices need a higher relative crystalline fraction todrive phase separation.37 However, the blend-cast device

efficiencies also correlate more strongly to the relativecrystalline fraction of the pure polymer than to that of theblend-cast polymer/fullerene composite. This indicates thatrather than studying the more complicated mixed systems usingX-ray diffraction, the pure polymer morphology is a betterpredictor of the final BHJ morphology and device performance.One surprising feature of Table 3 is the essentially complete

lack of correlation between (100) crystallite size and deviceefficiency for either blend-cast or sequentially processed RBOxdevices. Since large crystallites would prevent proper mixing,we expected that there would be an anticorrelation betweencrystalline domain size and device efficiency, which indeed wesee in the slow-dried blend-cast case. The degree of crystallinityis a combination of the crystallite size, the defect density withinthose crystallites, and the relative number of crystallites. Thus,the lack of correlation of device performance with crystallitesize for the other three fabrication conditions either indicatesthat as long as crystallites are below a certain size, reasonableperformance is observed, or more likely, that defects within acrystallite do not significantly affect performance, and so abroad range of apparent domain sizes can correlate with similarphysical crystallite sizes and thus similar device performance.In addition, although there is a correlation between

paracrystalline disorder and side-chain length, the para-crystallinity does not correlate with device efficiency. Thissuggests that disorder in the side-chains (rather than thesemiconducting polymer backbone) does not play a significantrole for the performance of devices fabricated by either SqP orblend-casting. We show in the SI that there is a strongcorrelation between the shape of the UV−visible absorptionspectrum and device performance, as the absorption spectrumaccurately reflects the type of order that is important for chargetransport through the BHJ device.67 We also show in the SI thecorrelation of relative fraction crystallinity with other deviceperformance figures of merit, including the short-circuitcurrent, open-circuit voltage, and fill factor. We find that the

Table 2. Summary of Pearson Product−Moment Correlation Coefficients between Different GIWAXS Parameters of the RBOxFilms Used in This Study and the RBOx Regioregularity (RR), Polydispersity Index (PDI), and Side-Chain Lengtha

correlation parameter (100) peak area (100) crystallite size g-factor

active layer RR PDI side-chain RR PDI side-chain RR PDI side-chain

slow dried SqP 0.68 0.22 0.24 −0.37 0.47 −0.89 0.45 0.08 0.79slow dried BC 0.74 0.72 −0.53 −0.59 0.34 −0.73 0.55 −0.17 0.84fast dried SqP 0.91 0.57 −0.14 −0.23 0.04 −0.83 0.03 −0.01 0.79fast dried BC −0.96 −0.36 −0.02 −0.55 −0.33 −0.61 0.63 −0.25 0.58

aAll SqP correlations are using pure polymer diffraction, not BHJs.

Table 3. Summary of Pearson Product−Moment CorrelationCoefficients between RBOx/PC71BM Device PowerConversion Efficiency and Different GIWAXS ParametersUsed in This Study

active layer(100)area

(010)area

(100)crystallite size

(010)crystallite size g-factor

slow driedSqP

−0.57a −0.35a 0.19a −0.92a −0.29a

slow driedBC

0.53a 0.21 −0.75 0.36 0.74

fast driedSqP

−0.96a 0.54a −0.01a −0.98a 0.15a

fast driedBC

−0.80 0.50 0.06 −0.85 0.20

aData compared to pure polymer films instead of BHJs.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

22120

relative fraction of crystallinity does not affect the open-circuitvoltage of the devices, a result consistent with drift−diffusionsimulations that have shown that VOC is relatively impervious todisorder and local environment.68 Instead, the correlationbetween relative crystalline fraction and device efficiency comesabout because of correlation with both the short-circuit currentand fill factor.C. Swelling Studies of Pure RBOx Films. Even though the

average side-chain length varies considerably across the RBOxseries, why are the two most important factors that correlatewith device efficiency the polymer regioregularity and therelative crystalline fraction in the pure film? We believe thatanswer is linked with how polymers of different regioregularityswell while being processing into BHJs. We have previouslyshown that polymer swelling is essential for producing high-efficiency photovoltaic devices fabricated through SqP, wherethe amorphous domains of the polymer are preferentiallyswollen.31 We also believe that swelling plays an important rolein the demixing of the polymer and fullerene during blend-casting, as discussed further below.To understand how differences in regioregularity and relative

crystalline fraction of the RBOx polymers are related toswelling, we measured the extent of swelling in both slow-driedand fast-dried RBOx films using spectroscopic ellipsometricporosimetry.51,69−74 For these studies, we induced swelling byexposing the polymer films to controlled amounts of toluenevapor, which is a good solvent for all of the polymers in thisseries. Figure 4a,b shows the resulting swelling profiles thatoccurred for each RBOx using fast-drying and slow-drying. Themaximum swelling of each film occurred at the saturation vaporpressure of toluene, 29 Torr. Figure 4c shows the maximumswelling of both fast-dried and slow-dried films of the purepolymers across the RBOx series, and Table 4 explores thecorrelations of swelling with various synthetic, structural, anddevice parameters. The swelling of both fast-dried and slow-dried RBOx films are anticorrelated to regioregularity, with themost regioregular polymers showing the lowest ability to swell.This result is expected, since highly regioregular polymersshould have a higher fraction relative crystallinity, leaving feweramorphous regions to uptake solvent.28,31,58,59,66,75−77 We alsosee an anticorrelation between swelling and PDI, which webelieve simply reflects the fact that, for this batch of samples,the more regioregular polymers also tended to have higherPDIs. We find little correlation between the average side-chainlength and the amount of swelling in the RBOx films. If weexpect swelling to be a major factor in controlling morphologyand thus BHJ device performance, this again indicates that side-chain length is not an important factor in determining thatmorphology.For the sequentially processed devices, we observe a strong

correlation between swelling and device efficiency for both theslow-dried and fast-dried polymer samples. This fits well withour previous work showing that swelling is the dominant factorthat controls fullerene intercalation, BHJ morphology, and thusthe performance of sequentially processed devices.31 Un-expectedly, we also see a modest anticorrelation betweenblend-cast device efficiency and pure polymer swelling in boththe fast-dried (−0.69) and slow-dried (−0.41) cases. Thisprovides strong evidence that swelling of a polymer film alsoplays a role in the kinetics of demixing that occurs duringdrying of blend-cast films. If the polymer is easily swollen, thereis less driving force to cause phase separation of the polymerand the fullerene, creating a different BHJ morphology than

would be obtained with a polymer that is difficult to swell. Thiscorrelation between demixing and swelling is part of whatcomplicates the X-ray diffraction observed in blend-cast mixed

Figure 4. Ellipsometric porosimetry swelling profiles for fast-dried (a)and slow-dried (b) pure RBOx films. (c) The maximum volumechange of swelled pure RBOx films taken at the saturation pressure oftoluene (29 Torr).

Table 4. Summary of Pearson Product−Moment CorrelationCoefficients between Pure RBOx Polymer Film Swelling atthe Saturation Pressure of Toluene with Different PolymerProperties, Device Efficiency, and the Various GIWAXSParameters Used in This Study

correlation to swelling slow-dried fast-dried

regioregularity −0.60 −0.81PDI −0.91 −0.63average side-chain length −0.21 −0.20SqP device efficiency 0.70 0.89BC device efficiency −0.41 −0.69SqP paracrystallinity −0.50 −0.01SqP (100) peak area −0.57 −0.47SqP (100) crystallite size −0.03 0.13SqP (010) peak area −0.37 0.26SqP (010) crystallite size −0.54 −0.84BC paracrystallinity −0.19 −0.54BC (100) peak area −0.56 0.47BC (100) crystallite size −0.02 0.48BC (010) peak area −0.26 0.19BC (010) crystallite size −0.20 0.45

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

22121

polymer/fullerene films, explaining why the best correlationsare to the fraction relative crystallinity of the pure polymerfilms. Thus, the crystallinity and swelling of pure polymersfilms, which are determined primarily by the polymerregioregularity, are much better predictors of the final blend-cast morphology than any parameter obtained through analysisof the BHJ film.

■ CONCLUSIONSThis work shows that it is the fundamental properties ofconjugated polymer chains determined by their synthesis, suchas the average chain regioregularity, that dominate BHJmorphologies produced by both SqP and blend-casting.Moreover, because SqP and blend-casting produce optimizedBHJ morphologies from fundamentally different starting points,factors such as polymer regioregularity control device perform-ance and morphology from these two processing routes inopposite ways. High polymer regioregularity is needed forblend-cast devices in order to drive phase separation betweenpolymers and fullerenes that were molecularly mixed in thesolution used to cast the film.3344,45,74−78 When high polymerregioregularity is not present, other processing techniques thatfacilitate demixing must be used, such as thermal annealing14,80

or the use of solvent additives11,79,80 or solvent annealing.72 InSqP, on the other hand, BHJ formation occurs only when thefullerene-casting solvent swells the underlying polymerfilm,8,23,31,37 a process that is facilitated with polymer samplesthat have lower regioregularity and relative fraction crystallinity.The initial goal of this work was to use polymer average side-

chain length to direct morphology formation in both blend-castand sequentially processed devices. What we found, however, isthat the regioregularity of the polymer, which happened to varyfrom one batch to the next, is the primary polymer propertyassociated with the overall BHJ morphology because of the wayregioregularity impacts crystallinity and swelling. We also foundthat the relative fraction crystallinity of pure polymer films is abetter predictor of device performance than any structuralcharacteristic of the BHJ blends. We note that thepolythiophene derivatives studied in this work, even thosewith lower regioregularity, tend to have higher relativecrystalline fractions, whereas low bandgap, high-efficiencypush−pull polymers tend to be much more amorphous. Thissuggests that regioregularity effects like those observed herecould be even more important with low bandgap materials thathave little fraction crystallinity and thus little driving force fordemixing with fullerenes in blend-cast devices; control overswelling of those materials is likely to be even more importantfor BHJ formation. This also suggests that the generally moreamorphous low-bandgap polymers should be more amenable toSqP than blend-casting, even though most of the work in theliterature with these materials has focused on blend-casting.Thus, although the correlations in this paper do not necessarilyprove causation, they do show that the regioregularity of apolymer batch is far more important in determining deviceproperties than is typically considered in the literature, and thatdifferent processing methods can be chosen to take advantageof the various polymer properties correlated with regioregu-larity.Finally, it is worth pointing out that the best sequentially

processed device from this series, fabricated with RBO72,outperforms the most efficient blend-cast device from thisseries, made with RBO50. This shows the danger in focusingsolely on the blend-cast fabrication technique to screen novel

photovoltaic materials. Although there is only a small differencein device efficiency between the best blend-cast and SqPdevices, these two processing methods favor different polymerswithin the RBOx series. What is particularly striking is that thebest material for one processing method is the other’s worstperformer. Therefore, knowledge gained by studying a materialsseries solely through the blend-casting method could bemisleading, particularly since the effects of regioregularity,swelling, and crystallinity run in opposite directions for blend-casting and SqP. For the case of the RBOx series, the highest-performing polymer RBO72 would have been entirely dismissedif only blend-cast films were studied. Clearly, it is necessary toconsider multiple fabrication methods when screening newmaterials for photovoltaic performance.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.jpcc.6b03300.

(PDF)

■ AUTHOR INFORMATIONCorresponding Authors*E-mail: tolbert@chem.ucla.edu; Tel: (310) 206-4767.*E-mail: schwartz@chem.ucla.edu; Tel: (310) 206-4113.Author Contributions⊥These authors contributed equally to this work.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by the National Science Foundation(NSF) under grants CHE-1112569, CHE-1608957 and CBET-1510353. J.C.A. acknowledges previous support from the NSFIGERT: Materials Creation Training Program (MCTP), grantnumber DGE-0654431 and the California NanoSystemsInstitute. A.S.F. acknowledges support from The Clean GreenIGERT (CGI), an energy-based NSF IGERT program (DGE-0903720). The X-ray diffraction studies presented in thismanuscript were carried out at the Stanford SynchrotronRadiation Lightsource. Use of the Stanford SynchrotronRadiation Lightsource, SLAC National Accelerator Laboratory,is supported by the U.S. Department of Energy, Office ofScience, Office of Basic Energy Sciences, under Contract DE-AC02-76SF00515. The authors wish to thank to MatthewFontana and Taylor Aubry for fabricating the ellipsometrysamples.

■ REFERENCES(1) Green, M. A.; Emery, K.; Hishikawa, Y.; Warta, W.; Dunlop, E. D.Solar Cell Efficiency Tables (version 43). Prog. Photovoltaics 2015, 23,1−9.(2) He, Z.; Zhong, C.; Su, S.; Xu, M.; Wu, H.; Cao, Y. EnhancedPower-Conversion Efficiency in Polymer Solar Cells Using an InvertedDevice Structure. Nat. Photonics 2012, 6, 593−597.(3) Collins, B. A.; Tumbleston, J. R.; Ade, H. Miscibility,Crystallinity, and Phase Development in P3HT/PCBM Solar Cells:Toward an Enlightened Understanding of Device Morphology andStability. J. Phys. Chem. Lett. 2011, 2, 3135−3145.(4) Shaw, P. E.; Ruseckas, A.; Samuel, I. D. W. Exciton DiffusionMeasurements in Poly(3-Hexylthiophene). Adv. Mater. 2008, 20,3516−3520.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

22122

(5) Groves, C.; Marsh, R. A.; Greenham, N. C. Monte CarloModeling of Geminate Recombination in Polymer-Polymer Photo-voltaic Devices. J. Chem. Phys. 2008, 129, 114903.(6) Halls, J. J. M.; Walsh, C. A.; Greenham, N.; Marseglia, E. A.;Friend, R.; Moratti, S. C.; Holmes, A. Efficient Photodiodes fromInterpenetrating Polymer Networks. Nature 1995, 376, 498−500.(7) Shaheen, S. E.; Brabec, C. J.; Sariciftci, N. S.; Padinger, F.;Fromherz, T.; Hummelen, J. C. 2.5% Efficient Organic Plastic SolarCells. Appl. Phys. Lett. 2001, 78, 841−843.(8) Yu, G.; Gao, J.; Hummelen, J. C.; Wudl, F.; Heeger, A. J. PolymerPhotovoltaic Cells: Enhanced Efficiencies via a Network of InternalDonor-Acceptor Heterojunctions. Science (Washington, DC, U. S.)1995, 270, 1789−1791.(9) Peet, J.; Soci, C.; Coffin, R. C.; Nguyen, T. Q.; Mikhailovsky, A.;Moses, D.; Bazan, G. C. Method for Increasing the PhotoconductiveResponse in Conjugated Polymer/fullerene Composites. Appl. Phys.Lett. 2006, 89, 252105.(10) Peet, J.; Kim, J. Y.; Coates, N. E.; Ma, W. L.; Moses, D.; Heeger,A. J.; Bazan, G. C. Efficiency Enhancement in Low-Bandgap PolymerSolar Cells by Processing with Alkane Dithiols. Nat. Mater. 2007, 6,497−500.(11) Lee, J. K.; Ma, W. L.; Brabec, C. J.; Yuen, J.; Moon, J. S.; Kim, J.Y.; Lee, K.; Bazan, G. C.; Heeger, A. J. Processing Additives forImproved Efficiency from Bulk Heterojunction Solar Cells. J. Am.Chem. Soc. 2008, 130, 3619−3623.(12) Ma, W.; Yang, C.; Gong, X.; Lee, K.; Heeger, A. J. ThermallyStable, Efficient Polymer Solar Cells with Nanoscale Control of theInterpenetrating Network Morphology. Adv. Funct. Mater. 2005, 15,1617−1622.(13) Kim, K.; Liu, J.; Namboothiry, M. A. G.; Carroll, D. L. Roles ofDonor and Acceptor Nanodomains in 6% Efficient ThermallyAnnealed Polymer Photovoltaics. Appl. Phys. Lett. 2007, 90, 163511.(14) Verploegen, E.; Mondal, R.; Bettinger, C. J.; Sok, S.; Toney, M.F.; Bao, Z. Effects of Thermal Annealing Upon the Morphology ofPolymer-Fullerene Blends. Adv. Funct. Mater. 2010, 20, 3519−3529.(15) Zhao, Y.; Xie, Z.; Qu, Y.; Geng, Y.; Wang, L. Solvent-VaporTreatment Induced Performance Enhancement of poly(3-Hexylth-iophene):methanofullerene Bulk-Heterojunction Photovoltaic Cells.Appl. Phys. Lett. 2007, 90, 043504.(16) Jo, J.; Na, S.-I.; Kim, S.-S.; Lee, T.-W.; Chung, Y.; Kang, S.-J.;Vak, D.; Kim, D.-Y. Three-Dimensional Bulk HeterojunctionMorphology for Achieving High Internal Quantum Efficiency inPolymer Solar Cells. Adv. Funct. Mater. 2009, 19, 2398−2406.(17) Heeger, A. J. 25th Anniversary Article: Bulk HeterojunctionSolar Cells: Understanding the Mechanism of Operation. Adv. Mater.2014, 26, 10−28.(18) Ayzner, A. L.; Tassone, C. J.; Tolbert, S. H.; Schwartz, B. J.Reappraising the Need for Bulk Heterojunctions in Polymer/FullerenePhotovoltaics: The Role of Carrier Transport in All-Solution-Processed P3HT/PCBM Bilayer Solar Cells. J. Phys. Chem. C 2009,113, 20050−20060.(19) Gevaerts, V. S.; Koster, L. J. A.; Wienk, M. M.; Janssen, R. A. J.Discriminating between Bilayer and Bulk Heterojunction Polymer:-Fullerene Solar Cells Using the External Quantum Efficiency. ACSAppl. Mater. Interfaces 2011, 3, 3252−3255.(20) Moon, J. S.; Takacs, C. J.; Sun, Y.; Heeger, A. J. SpontaneousFormation of Bulk Heterojunction Nanostructures: Multiple Routes toEquivalent Morphologies. Nano Lett. 2011, 11, 1036−1039.(21) Treat, N. D.; Brady, M. A.; Smith, G.; Toney, M. F.; Kramer, E.J.; Hawker, C. J.; Chabinyc, M. L. Interdiffusion of PCBM and P3HTReveals Miscibility in a Photovoltaically Active Blend. Adv. EnergyMater. 2011, 1, 82−89.(22) Ayzner, A. L.; Doan, S. C.; Tremolet de Villers, B.; Schwartz, B.J. Ultrafast Studies of Exciton Migration and Polaron Formation inSequentially Solution-Processed Conjugated Polymer/FullereneQuasi-Bilayer Photovoltaics. J. Phys. Chem. Lett. 2012, 3, 2281−2287.(23) Hawks, S. A.; Aguirre, J. C.; Schelhas, L. T.; Thompson, R. J.;Huber, R. C.; Ferreira, A. S.; Zhang, G.; Herzing, A. A.; Tolbert, S. H.;Schwartz, B. J. Comparing Matched Polymer:Fullerene Solar Cells

Made by Solution-Sequential Processing and Traditional BlendCasting: Nanoscale Structure and Device Performance. J. Phys.Chem. C 2014, 118, 17413−17425.(24) Zhang, G.; Huber, R.; Ferreira, A.; Boyd, S. D.; Luscombe, C.K.; Tolbert, S. H.; Schwartz, B. J. Crystallinity Effect in SequentiallyProcessed and Blend Cast Bulk-Heterojunction Polymer/FullerenePhotovoltaics. J. Phys. Chem. C 2014, 118, 18424−18435.(25) Clark, J.; Silva, C.; Friend, R.; Spano, F. Role of IntermolecularCoupling in the Photophysics of Disordered Organic Semiconductors:Aggregate Emission in Regioregular Polythiophene. Phys. Rev. Lett.2007, 98, 206406−206406−4.(26) Van Franeker, J. J.; Kouijzer, S.; Lou, X.; Turbiez, M.; Wienk, M.M.; Janssen, R. A. J. Depositing Fullerenes in Swollen Polymer Layersvia Sequential Processing of Organic Solar Cells. Adv. Energy Mater.2015, 5, 1500464.(27) Lee, K. H.; Schwenn, P. E.; Smith, A. R. G.; Cavaye, H.; Shaw, P.E.; James, M.; Krueger, K. B.; Gentle, I. R.; Meredith, P.; Burn, P. L.Morphology of All-Solution-Processed “Bilayer” Organic Solar Cells.Adv. Mater. 2011, 23, 766−770.(28) Loiudice, A.; Rizzo, A.; Latini, G.; Nobile, C.; de Giorgi, M.;Gigli, G. Graded Vertical Phase Separation of Donor/acceptor Speciesfor Polymer Solar Cells. Sol. Energy Mater. Sol. Cells 2012, 100, 147−152.(29) Nardes, A. M.; Ayzner, A. L.; Hammond, S. R.; Ferguson, A. J.;Schwartz, B. J.; Kopidakis, N. Photoinduced Charge CarrierGeneration and Decay in Sequentially Deposited Polymer/FullereneLayers: Bulk Heterojunction vs Planar Interface. J. Phys. Chem. C 2012,116, 7293−7305.(30) Kim, D. H.; Mei, J.; Ayzner, A. L.; Schmidt, K.; Giri, G.;Appleton, A. L.; Toney, M. F.; Bao, Z. Sequentially Solution-Processed, Nanostructured Polymer Photovoltaics Using SelectiveSolvents. Energy Environ. Sci. 2014, 7, 1103−1109.(31) Aguirre, J. C.; Hawks, S. A.; Ferreira, A.; Yee, P.; Subramaniyan,S.; Jenekhe, S. A.; Tolbert, S. H.; Schwartz, B. J. Sequential Processingfor Organic Photovoltaics: Design Rules for Morphology Control byTailored Semi-Orthogonal Solvent Blends. Adv. Energy Mater. 2015, 5,1402020.(32) Liu, Y.; Liu, F.; Wang, H. W.; Nordlund, D.; Sun, Z.; Ferdous,S.; Russell, T. P. Sequential Deposition: Optimization of SolventSwelling for High-Performance Polymer Solar Cells. ACS Appl. Mater.Interfaces 2015, 7, 653−661.(33) Collins, B. A.; Gann, E.; Guignard, L.; He, X.; McNeill, C. R.;Ade, H. Molecular Miscibility of Polymer−Fullerene Blends. J. Phys.Chem. Lett. 2010, 1, 3160−3166.(34) Lee, K. H.; Zhang, Y.; Burn, P. L.; Gentle, I. R.; James, M.;Nelson, A.; Meredith, P. Correlation of Diffusion and Performance inSequentially Processed P3HT/PCBM Heterojunction Films by Time-Resolved Neutron Reflectometry. J. Mater. Chem. C 2013, 1, 2593.(35) Aguirre, J. C.; Ferreira, A.; Ding, H.; Jenekhe, S. A.; Kopidakis,N.; Asta, M.; Pilon, L.; Rubin, Y.; Tolbert, S. H.; Schwartz, B. J.; et al.Panoramic View of Electrochemical Pseudocapacitor and OrganicSolar Cell Research in Molecularly Engineered Energy Materials(MEEM). J. Phys. Chem. C 2014, 118, 19505−19523.(36) Babel, A.; Jenekhe, S. a. Alkyl Chain Length Dependence of theField-Effect Carrier Mobility in Regioregular poly(3-Alkylthiophene)s.Synth. Met. 2005, 148, 169−173.(37) Gadisa, A.; Oosterbaan, W. D.; Vandewal, K.; Bolsee, J. C.;Bertho, S.; D’Haen, J.; Lutsen, L.; Vanderzande, D.; Manca, J. V. Effectof Alkyl Side-Chain Length on Photovoltaic Properties of poly(3-alkylthiophene)/PCBM Bulk Heterojunctions. Adv. Funct. Mater.2009, 19, 3300−3306.(38) Shen, X.; Hu, W.; Russell, T. P. Measuring the Degree ofCrystallinity in Semicrystalline Regioregular Poly(3-Hexylthiophene).Macromolecules 2016, 49, 4501.(39) Wu, P. T.; Ren, G.; Jenekhe, S. A. Crystalline RandomConjugated Copolymers with Multiple Side Chains: TunableIntermolecular Interaction and Enhanced Charge Transport andPhotovoltaic Properties. Macromolecules 2010, 43, 3306−3313.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

22123

(40) Burkhart, B.; Khlyabich, P. P.; Thompson, B. C. Influence of theEthylhexyl Side-Chain Content on the Open-Circuit Voltage in Rr-Poly(3-Hexylthiophene- Co −3-(2-Ethylhexyl)thiophene) Copoly-mers. Macromolecules 2012, 45, 3740−3748.(41) Burkhart, B.; Khlyabich, P. P.; Cakir Canak, T.; LaJoie, T. W.;Thompson, B. C. Semi-Random” Multichromophoric Rr-P3HTAnalogues for Solar Photon Harvesting. Macromolecules 2011, 44,1242−1246.(42) Gao, D.; Hollinger, J.; Jahnke, A. A.; Seferos, D. S. Influence ofSelenophene−thiophene Phase Separation on Solar Cell Performance.J. Mater. Chem. A 2014, 2, 6058−6063.(43) Yan, H.; Hollinger, J.; Bridges, C. R.; McKeown, G. R.; Al-Faouri, T.; Seferos, D. S. Doping Poly(3-Hexylthiophene) Nanowireswith Selenophene Increases the Performance of Polymer-NanowireSolar Cells. Chem. Mater. 2014, 26, 4605−4611.(44) Kim, Y.; Cook, S.; Tuladhar, S. M.; Choulis, S. A.; Nelson, J.;Durrant, J. R.; Bradley, D. D. C.; Giles, M.; McCulloch, I.; Ha, C.-S.;et al. A Strong Regioregularity Effect in Self-Organizing ConjugatedPolymer Films and High-Efficiency Polythiophene:fullerene SolarCells. Nat. Mater. 2006, 5, 197−203.(45) Mauer, R.; Kastler, M.; Laquai, F. The Impact of PolymerRegioregularity on Charge Transport and Efficiency of P3HT:PCBMPhotovoltaic Devices. Adv. Funct. Mater. 2010, 20, 2085−2092.(46) Woo, C. H.; Thompson, B. C.; Kim, B. J.; Toney, M. F.;Frechet, J. M. J. The Influence of poly(3-Hexylthiophene)Regioregularity on Fullerene-Composite Solar Cell Performance. J.Am. Chem. Soc. 2008, 130, 16324−16329.(47) Sivula, K.; Luscombe, C. K.; Thompson, B. C.; Frechet, J. M. J.Enhancing the Thermal Stability of Polythiophene:fullerene SolarCells by Decreasing Effective Polymer Regioregularity. J. Am. Chem.Soc. 2006, 128, 13988−13989.(48) Sher, M.-J.; Bartelt, J. A.; Burke, T. M.; Salleo, A.; McGehee, M.D.; Lindenberg, A. M. Time- and Temperature-Independent LocalCarrier Mobility and Effects of Regioregularity in Polymer-FullereneOrganic Semiconductors. Adv. Electron. Mater. 2016, 2, 1500351.(49) Snyder, C. R.; Henry, J. S.; Delongchamp, D. M. Effect ofRegioregularity on the Semicrystalline Structure of poly(3-Hexylth-iophene). Macromolecules 2011, 44, 7088−7091.(50) Lilliu, S.; Agostinelli, T.; Pires, E.; Hampton, M.; Nelson, J.;Macdonald, J. E. Dynamics of Crystallization and Disorder duringAnnealing of P3HT/PCBM Bulk Heterojunctions. Macromolecules2011, 44, 2725−2734.(51) Huttner, S.; Sommer, M.; Chiche, A.; Krausch, G.; Steiner, U.;Thelakkat, M. Controlled Solvent Vapour Annealing for PolymerElectronics. Soft Matter 2009, 5, 4206−4211.(52) Ng, A.; Li, C. H.; Fung, M. K.; Djurisic, A. B.; Zapien, J. A.;Chan, W. K.; Cheung, K. Y.; Wong, W.-Y. Accurate Determination ofthe Index of Refraction of Polymer Blend Films by SpectroscopicEllipsometry. J. Phys. Chem. C 2010, 114, 15094−15101.(53) Hedley, G. J.; Ward, A. J.; Alekseev, A.; Howells, C. T.; Martins,E. R.; Serrano, L. A.; Cooke, G.; Ruseckas, A.; Samuel, I. D. W.Determining the Optimum Morphology in High-PerformancePolymer-Fullerene Organic Photovoltaic Cells. Nat. Commun. 2013,4, 2867.(54) Pearson, K. Note on Regression and Inheritance in the Case ofTwo Parents. Proc. R. Soc. London 1895, 58, 240−242.(55) Jiang, X. M.; Osterbacka, R.; Korovyanko, O.; An, C. P.;Horovitz, B.; Janssen, R. A. J.; Vardeny, Z. V. Spectroscopic Studies ofPhotoexcitations in Regioregular and Regiorandom PolythiopheneFilms. Adv. Funct. Mater. 2002, 12, 587−597.(56) Adachi, T.; Brazard, J.; Ono, R. J.; Hanson, B.; Traub, M. C.;Wu, Z.-Q.; Li, Z.; Bolinger, J. C.; Ganesan, V.; Bielawski, C. W.; et al.Regioregularity and Single Polythiophene Chain Conformation. J.Phys. Chem. Lett. 2011, 2, 1400−1404.(57) Kohn, P.; Rong, Z.; Scherer, K. H.; Sepe, A.; Sommer, M.;Muller-Buschbaum, P.; Friend, R. H.; Steiner, U.; Huttner, S.Crystallization-Induced 10-Nm Structure Formation in P3HT/PCBM Blends. Macromolecules 2013, 46, 4002−4013.

(58) Turner, S. T.; Pingel, P.; Steyrleuthner, R.; Crossland, E. J. W.;Ludwigs, S.; Neher, D. Quantitative Analysis of Bulk HeterojunctionFilms Using Linear Absorption Spectroscopy and Solar CellPerformance. Adv. Funct. Mater. 2011, 21, 4640−4652.(59) Rogers, J. T.; Schmidt, K.; Toney, M. F.; Kramer, E. J.; Bazan, G.C. Structural Order in Bulk Heterojunction Films Prepared withSolvent Additives. Adv. Mater. 2011, 23, 2284−2288.(60) Verploegen, E.; Miller, C. E.; Schmidt, K.; Bao, Z.; Toney, M. F.Manipulating the Morphology of P3HT−PCBM Bulk HeterojunctionBlends with Solvent Vapor Annealing. Chem. Mater. 2012, 24, 3923−3931.(61) Poelking, C.; Andrienko, D. Effect of Polymorphism,Regioregularity and Paracrystallinity on Charge Transport in Poly(3-Hexylthiophene) [P3HT] Nanofibers. Macromolecules 2013, 46,8941−8956.(62) Lilliu, S.; Agostinelli, T.; Verploegen, E.; Pires, E.; Hampton, M.;Al-Hashimi, M.; Heeney, M. J.; Toney, M. F.; Nelson, J.; Macdonald, J.E. Effects of Thermal Annealing upon the Nanomorphology of poly(3-Hexylselenophene)-PCBM Blends. Macromol. Rapid Commun. 2011,32, 1454−1460.(63) Noriega, R.; Rivnay, J.; Vandewal, K.; Koch, F. P. V; Stingelin,N.; Smith, P.; Toney, M. F.; Salleo, A. A General Relationship betweenDisorder, Aggregation and Charge Transport in Conjugated Polymers.Nat. Mater. 2013, 12, 1038.(64) Sanyal, M.; Schmidt-Hansberg, B.; Klein, M. F. G.; Colsmann,A.; Munuera, C.; Vorobiev, A.; Lemmer, U.; Schabel, W.; Dosch, H.;Barrena, E. In Situ X-Ray Study of Drying-Temperature Influence onthe Structural Evolution of Bulk-Heterojunction Polymer−FullereneSolar Cells Processed by Doctor-Blading. Adv. Energy Mater. 2011, 1,363−367.(65) Shin, N.; Richter, L. J.; Herzing, A. A.; Kline, R. J.;DeLongchamp, D. M. Effect of Processing Additives on theSolidification of Blade-Coated Polymer/Fullerene Blend Films viaIn-Situ Structure Measurements. Adv. Energy Mater. 2013, 3, 938−948.(66) Zhang, G.; Ferreira, A.; Huber, R.; Boyd, S. D.; Luscombe, C.K.; Tolbert, S. H.; Schwartz, B. J. Crystallinity Effect in Sequentially-Processed P3HT/PCBM Photovoltaics. Submitted for publication.(67) Spano, F. C. Modeling Disorder in Polymer Aggregates: TheOptical Spectroscopy of Regioregular poly(3-Hexylthiophene) ThinFilms. J. Chem. Phys. 2005, 122, 234701−234715.(68) Finck, B. Y.; Schwartz, B. J. Drift-Diffusion Modeling of theEffects of Structural Disorder and Carrier Mobility on the Performanceof Organic Photovoltaic Devices. Phys. Rev. Appl. 2015, 4, 034006.(69) Elbs, H.; Krausch, G. Ellipsometric Determination of Flory-Huggins Interaction Parameters in Solution. Polymer 2004, 45, 7935−7942.(70) Ogieglo, W.; van der Werf, H.; Tempelman, K.; Wormeester,H.; Wessling, M.; Nijmeijer, A.; Benes, N. E. N-Hexane InducedSwelling of Thin PDMS Films under Non-Equilibrium NanofiltrationPermeation Conditions, Resolved by Spectroscopic Ellipsometry. J.Membr. Sci. 2013, 437, 313−323.(71) Papanu, J. S.; Hess, D. W.; Soane, D. S.; Bell, A. T. Dissolutionof Thin Poly(methyl Methacrylate) Films in Ketones, Binary Ketone/Alcohol Mixtures, and Hydroxy Ketones. J. Electrochem. Soc. 1989, 136,3077−3083.(72) Papanu, J. S.; Hess, D. W.; Soane (Soong), D. S.; Bell, A. T.Swelling of Poly(methyl Methacrylate) Thin Films in Low MolecularWeight Alcohols. J. Appl. Polym. Sci. 1990, 39, 803−823.(73) Sirard, S. M.; Green, P. F.; Johnston, K. P. SpectroscopicEllipsometry Investigation of the Swelling of Poly(Dimethylsiloxane)Thin Films with High Pressure Carbon Dioxide. J. Phys. Chem. B 2001,105, 766−772.(74) Stamatialis, D. F.; Sanopoulou, M.; Raptis, I. Swelling andDissolution Behavior of Poly(methyl Methacrylate) Films in MethylEthyl Ketone/methyl Alcohol Mixtures Studied by Optical Techni-ques. J. Appl. Polym. Sci. 2002, 83, 2823−2834.(75) Chen, D.; Liu, F.; Wang, C.; Nakahara, A.; Russell, T. P. BulkHeterojunction Photovoltaic Active Layers via Bilayer Interdiffusion.Nano Lett. 2011, 11, 2071−2078.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

22124

(76) Vohra, V.; Arrighetti, G.; Barba, L.; Higashimine, K.; Porzio, W.;Murata, H. Enhanced Vertical Concentration Gradient in RubbedP3HT:PCBM Graded Bilayer Solar Cells. J. Phys. Chem. Lett. 2012, 3,1820−1823.(77) Vohra, V.; Higashimine, K.; Murakami, T.; Murata, H. Additionof Regiorandom Poly (3-Hexylthiophene) to Solution ProcessedGraded Bilayers to Tune the Vertical Concentration Gradient. Appl.Phys. Lett. 2012, 101, 173301.(78) Chiu, M.-Y.; Jeng, U.-S.; Su, M.-S.; Wei, K.-H. Morphologies ofSelf-Organizing Regioregular Conjugated Polymer/Fullerene Aggre-gates in Thin Film Solar Cells. Macromolecules 2010, 43, 428−432.(79) Lou, S. J.; Szarko, J. M.; Xu, T.; Yu, L.; Marks, T. J.; Chen, L. X.Effects of Additives on the Morphology of Solution Phase AggregatesFormed by Active Layer Components of High-Efficiency OrganicSolar Cells. J. Am. Chem. Soc. 2011, 133, 20661−20663.(80) Tang, H.; Lu, G.; Li, L.; Li, J.; Wang, Y.; Yang, X. PreciseConstruction of PCBM Aggregates for Polymer Solar Cells via Multi-Step Controlled Solvent Vapor Annealing. J. Mater. Chem. 2010, 20,683−688.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b03300J. Phys. Chem. C 2016, 120, 22115−22125

22125